amitmurthy · May 30, 2017 15:51
diff --git a/pmap_bench.jl b/pmap_bench.jl
 #using BenchmarkTools
 using NamedArrays

 nprocs()==1 && addprocs(4);

 data_szs = [10^n for n in 0:6];
 coll_lengths = [100, 1000, 10000];

 for (idx, sz) in enumerate(data_szs)
    sym = Symbol("gv_$idx")
    @eval global const $sym = ones($sz)

    fsym = Symbol("f_glbdata_$idx")
    @eval @everywhere $fsym(x) = length($sym)

    pmapsym = Symbol("pmap_glbdata_$idx")
    @eval $pmapsym(N) = pmap(x->length($sym), 1:N)
 end

 function make_results_array()
    results = NamedArray(fill(NaN, (length(data_szs),length(coll_lengths))))
    setnames!(results, [string("10^", ceil(Int, log(10, x)), " element Array{Float64,1}") for x in data_szs], 1)
    setdimnames!(results, "pmap collection length", 1)

    setnames!(results, [string(x) for x in coll_lengths], 2)
    setdimnames!(results, "input num elements", 2)
    results
 end

 function exec_block(results, f, warmup=false)
    for (idx_coll, N) in enumerate(coll_lengths)
        skip = false
        for (idx_dsz, sz) in enumerate(data_szs)
            if skip
                println("Skipping data_sz = $sz, coll_length = $N")
                continue
            end

            t = f(idx_coll, N, idx_dsz, sz)

            # if t > 2.0
            #     skip = true
            # end

            println("Executed data_sz = $sz, coll_length = $N in $t seconds")
            results[idx_dsz, idx_coll] = trunc(t,4)
            @everywhere gc()
            @everywhere gc()
        end
        warmup && break
    end
 end

 function run_glbdata_closure(idx_coll, N, idx_dsz, sz)
    fsym = Symbol("pmap_glbdata_$idx_dsz")
    f = getfield(Main, fsym)
    @elapsed f(N)
 end

 function run_glbdata_funcs(idx_coll, N, idx_dsz, sz)
    fsym = Symbol("f_glbdata_$idx_dsz")
    f = getfield(Main, fsym)
    @elapsed pmap(f, 1:N)
 end

 function run_closures(idx_coll, N, idx_dsz, sz)
    a = ones(sz)
    @elapsed pmap(x->length(a), 1:N)
 end



 function run_cp(idx_coll, N, idx_dsz, sz)
    a = ones(sz)
    cp = CachingPool(workers())
    t = @elapsed pmap(cp, x->length(a), 1:N)
    clear!(cp)
    t
 end

 const gcp = CachingPool(workers())
 function run_cp_reuse(idx_coll, N, idx_dsz, sz)
    a = ones(sz)
    @elapsed pmap(gcp, x->length(a), 1:N)
 end

 const cccp = CachingPool(workers())
 function run_cached_closure(idx_coll, N, idx_dsz, sz)
    a = ones(sz)
    cl = x->length(a)
    pmap(cccp, cl, 1:nworkers())  # cache cl in cccp
    t = @elapsed pmap(cccp, cl, 1:N)
 end


 function run()
    all_results=[]

    def_cleanup()=nothing
    for (hdr, f, cleanup) in [
                ("Regular pmap with closures pointing to global data", run_glbdata_closure, def_cleanup),
                # run_glbdata_closure first as it creates globals on all workers, referenced again by run_glbdata_funcs
                ("Regular pmap with defined functions pointing to global data", run_glbdata_funcs, def_cleanup),
                ("Regular pmap with local data captured in a closure", run_closures, def_cleanup),
                ("pmap with a caching pool created for each call. Using closures", run_cp, def_cleanup),
 #                ("pmap with a caching pool used across all calls. closures defined in each pmap call", run_cp_reuse, ()->clear!(gcp)),
                ("pmap with a caching pool used across all calls. Time cached closures call.", run_cached_closure, ()->clear!(cccp))
            ]

        println("\n", "Warmup run : ", hdr)
        results = make_results_array();
        exec_block(results, f, true) # warmup run
        println("\n", "Timed run : ", hdr)
        exec_block(results, f)
        push!(all_results, (hdr, results))
        cleanup()
    end
    all_results
 end


 for (hdr, results) in run()
    println(hdr, ":")
    @show results
    println()
 end
	#using BenchmarkTools
	using NamedArrays

	nprocs()==1 && addprocs(4);

	data_szs = [10^n for n in 0:6];
	coll_lengths = [100, 1000, 10000];

	for (idx, sz) in enumerate(data_szs)
	sym = Symbol("gv_$idx")
	@eval global const $sym = ones($sz)

	fsym = Symbol("f_glbdata_$idx")
	@eval @everywhere $fsym(x) = length($sym)

	pmapsym = Symbol("pmap_glbdata_$idx")
	@eval $pmapsym(N) = pmap(x->length($sym), 1:N)
	end

	function make_results_array()
	results = NamedArray(fill(NaN, (length(data_szs),length(coll_lengths))))
	setnames!(results, [string("10^", ceil(Int, log(10, x)), " element Array{Float64,1}") for x in data_szs], 1)
	setdimnames!(results, "pmap collection length", 1)

	setnames!(results, [string(x) for x in coll_lengths], 2)
	setdimnames!(results, "input num elements", 2)
	results
	end

	function exec_block(results, f, warmup=false)
	for (idx_coll, N) in enumerate(coll_lengths)
	skip = false
	for (idx_dsz, sz) in enumerate(data_szs)
	if skip
	println("Skipping data_sz = $sz, coll_length = $N")
	continue
	end

	t = f(idx_coll, N, idx_dsz, sz)

	# if t > 2.0
	# skip = true
	# end

	println("Executed data_sz = $sz, coll_length = $N in $t seconds")
	results[idx_dsz, idx_coll] = trunc(t,4)
	@everywhere gc()
	@everywhere gc()
	end
	warmup && break
	end
	end

	function run_glbdata_closure(idx_coll, N, idx_dsz, sz)
	fsym = Symbol("pmap_glbdata_$idx_dsz")
	f = getfield(Main, fsym)
	@elapsed f(N)
	end

	function run_glbdata_funcs(idx_coll, N, idx_dsz, sz)
	fsym = Symbol("f_glbdata_$idx_dsz")
	f = getfield(Main, fsym)
	@elapsed pmap(f, 1:N)
	end

	function run_closures(idx_coll, N, idx_dsz, sz)
	a = ones(sz)
	@elapsed pmap(x->length(a), 1:N)
	end



	function run_cp(idx_coll, N, idx_dsz, sz)
	a = ones(sz)
	cp = CachingPool(workers())
	t = @elapsed pmap(cp, x->length(a), 1:N)
	clear!(cp)
	t
	end

	const gcp = CachingPool(workers())
	function run_cp_reuse(idx_coll, N, idx_dsz, sz)
	a = ones(sz)
	@elapsed pmap(gcp, x->length(a), 1:N)
	end

	const cccp = CachingPool(workers())
	function run_cached_closure(idx_coll, N, idx_dsz, sz)
	a = ones(sz)
	cl = x->length(a)
	pmap(cccp, cl, 1:nworkers()) # cache cl in cccp
	t = @elapsed pmap(cccp, cl, 1:N)
	end


	function run()
	all_results=[]

	def_cleanup()=nothing
	for (hdr, f, cleanup) in [
	("Regular pmap with closures pointing to global data", run_glbdata_closure, def_cleanup),
	# run_glbdata_closure first as it creates globals on all workers, referenced again by run_glbdata_funcs
	("Regular pmap with defined functions pointing to global data", run_glbdata_funcs, def_cleanup),
	("Regular pmap with local data captured in a closure", run_closures, def_cleanup),
	("pmap with a caching pool created for each call. Using closures", run_cp, def_cleanup),
	# ("pmap with a caching pool used across all calls. closures defined in each pmap call", run_cp_reuse, ()->clear!(gcp)),
	("pmap with a caching pool used across all calls. Time cached closures call.", run_cached_closure, ()->clear!(cccp))
	]

	println("\n", "Warmup run : ", hdr)
	results = make_results_array();
	exec_block(results, f, true) # warmup run
	println("\n", "Timed run : ", hdr)
	exec_block(results, f)
	push!(all_results, (hdr, results))
	cleanup()
	end
	all_results
	end


	for (hdr, results) in run()
	println(hdr, ":")
	@show results
	println()
	end